Could you tell us a bit about how you approach the development of new features at Silverfin, particularly AI features?

Our approach has changed quite a few times over the past two years, because we keep challenging ourselves to find out what works best for us. We are probably a textbook example of a startup with a visionary founder that transitioned (and still is transitioning) into a more mature product organisation. Our current approach is to start from business outcomes, and have product trios (consisting of a product manager, product designer and engineer) identify opportunities that would contribute to reaching those outcomes by analysing user feedback, user behaviour, and most importantly talking to them. Depending on the estimated likelihood AI could play part in what we want to achieve, that product trio is extended to also include an AI engineer from the very beginning. 

We then align with business stakeholders on which opportunities to pursue, after which we validate our potential solutions again with users, typically by means of a clickable prototype or proof of concept implementation. For most AI features we’ll build a working proof of concept early on more often, because faking delays and accuracy in prototypes does not give the same experience. This also helps us de-risk technical feasibility early on. 

We will then re-assess whether we still think the opportunity is worth pursuing and whether our approach was right. If it was, we’ll break the solution down in iterations, aiming to ship small changes so we can keep validating actual user behaviour and feedback. If it wasn’t, we consider other solutions or pivot to another opportunity.

‍

What made you want to try a Design Sprint for developing your most recent AI feature?

As mentioned, we like to continuously challenge ourselves and our methods. Because of that, doing a design sprint was an obvious thing to do as soon as we learned Visma had a team facilitating them.

It was a great experience for us! It’s incredible how far you can get in just a week if you plan ahead and make it your sole focus for an entire team.

‍

How would you describe the impact of the Design Sprint on your teams and customers so far? 

When we did this design sprint, having engineers join from day one of discovery was still relatively new to our organisation. By now, that has become our standard way of doing things, but at that time it really helped our team see the value. The shared understanding you build makes collaboration so much easier.

We released the feature we worked on during the design sprint to all users in September last year, although it’s worth noting that our AI features are currently still a paid add-on.

The adoption of the feature was almost instantaneous. We saw that 80% of all users that used it did so again within a week.

‍

What are some successes and challenges you have experienced with this Design Sprint? For instance compared to your typical way of working.

It was challenging to entirely clear the calendars of the team participating in the design sprint for an entire week, but that was also one of the biggest drivers of this being a successful format.

The time it takes to make decisions and act on them is usually determined by when meetings with the team and stakeholders can be scheduled to align, so having them fully available allows you to move so much quicker. 

‍

Would you recommend Design Sprints to other companies that are introducing AI to their products? What advice would you give them?

Definitely. The Product Discovery team in Visma Group is clearly very experienced, both in facilitating the process and in making killer prototypes for applications they have just heard about. To us this was an opportunity to learn from them. I would not just recommend running a design sprint when you’re introducing AI features, but in general as well.

However, if it is your first time introducing AI to your product, you can be assured that the team facilitating your design sprint has done this before, and knows which nuances to consider when you’re validating an AI opportunity.

‍